OBJECTIVES: Lipid modifying agents (7.63 percent of total drug expenditure in 2009) were prescribed by several specialists in a Thai teaching hospital. Cluster analysis was applied to group prescribing physicians by drug cost. Pattern of prescription among clusters were studied.  METHODS: Data of drugs in Anatomical Therapeutic Chemical class C10 (lipid modifying agents) were retrieved from a hospital database, and unidentified physician records (1,758 from 205,964) were deleted. Drug cost for each prescribing physician was the summation of C10 drug issued quantity multiplied by unit cost at selling price. Drug costs for each physician were then grouped by hierarchical clustering. Pattern of prescribed drug among clusters was analyzed after classifying lipid lowering agents into 4 types i.e., brand statins, brand non-statins, generic statins, and generic non-statins.  RESULTS: Prescribing physicians were grouped into 4 clusters; one gynecologist in the first; two internists in the second; four internists, one family medicine doctor, one gynecologist in the third and the rest in the fourth cluster. Pattern of C10 drugs was 76.2 % brand statins, 19.1 % brand non-statins, 3.3 % generic statins, and 1.4 % non-statins. In the first cluster, 97.3 % of the 13.3 million Baht was from brand stains. For the second cluster, 9.15 and 9.88 million Baht was each prescribed by two doctors with 65.7 % from brand statins and 32.2 % from brand non-statins.  Distribution of drug type in the third cluster was similar to that in the fourth. However, average drug expenditure was higher in the third cluster than the fourth (6.11 million vs. 0.14 million Baht).  CONCLUSIONS: Patterns of prescribing behavior of lipid modifying agents can be identified through cluster analysis of prescription database. The result can be helpful for the further study of factors accountable for marked deviation from common patterns.